Upgrade to Aurora 0.22.0 (#5)
* Upgrading to Thrift 0.13.1. This version is a fork of 0.13.0 with a patch on top of it to fix an issue where trying a realis call after the connection has been closed results in a panic. * Upgrading compose set up to Mesos 1.6.2 and Aurora 0.22.0. * Adding support for using different update strategies. * Adding a monitor that is friendly with auto pause. * Adding tests for new update strategies.
This commit is contained in:
parent
1d8afcd329
commit
69ced895e2
15 changed files with 2013 additions and 1068 deletions
|
@ -716,9 +716,40 @@ struct JobUpdateKey {
|
|||
2: string id
|
||||
}
|
||||
|
||||
/** Limits the amount of active changes being made to instances to groupSize. */
|
||||
struct QueueJobUpdateStrategy {
|
||||
1: i32 groupSize
|
||||
}
|
||||
|
||||
/** Similar to Queue strategy but will not start a new group until all instances in an active
|
||||
* group have finished updating.
|
||||
*/
|
||||
struct BatchJobUpdateStrategy {
|
||||
1: i32 groupSize
|
||||
/* Update will pause automatically after each batch completes */
|
||||
2: bool autopauseAfterBatch
|
||||
}
|
||||
|
||||
/** Same as Batch strategy but each time an active group completes, the size of the next active
|
||||
* group may change.
|
||||
*/
|
||||
struct VariableBatchJobUpdateStrategy {
|
||||
1: list<i32> groupSizes
|
||||
/* Update will pause automatically after each batch completes */
|
||||
2: bool autopauseAfterBatch
|
||||
}
|
||||
|
||||
union JobUpdateStrategy {
|
||||
1: QueueJobUpdateStrategy queueStrategy
|
||||
2: BatchJobUpdateStrategy batchStrategy
|
||||
3: VariableBatchJobUpdateStrategy varBatchStrategy
|
||||
}
|
||||
|
||||
/** Job update thresholds and limits. */
|
||||
struct JobUpdateSettings {
|
||||
/** Max number of instances being updated at any given moment. */
|
||||
/** Deprecated, please set value inside of desired update strategy instead.
|
||||
* Max number of instances being updated at any given moment.
|
||||
*/
|
||||
1: i32 updateGroupSize
|
||||
|
||||
/** Max number of instance failures to tolerate before marking instance as FAILED. */
|
||||
|
@ -736,7 +767,7 @@ struct JobUpdateSettings {
|
|||
/** Instance IDs to act on. All instances will be affected if this is not set. */
|
||||
7: set<Range> updateOnlyTheseInstances
|
||||
|
||||
/**
|
||||
/** Deprecated, please set updateStrategy to the Batch strategy instead.
|
||||
* If true, use updateGroupSize as strict batching boundaries, and avoid proceeding to another
|
||||
* batch until the preceding batch finishes updating.
|
||||
*/
|
||||
|
@ -755,6 +786,9 @@ struct JobUpdateSettings {
|
|||
* differs between the old and new task configurations, updates will use the newest configuration.
|
||||
*/
|
||||
10: optional bool slaAware
|
||||
|
||||
/** Update strategy to be used for the update. See JobUpdateStrategy for choices. */
|
||||
11: optional JobUpdateStrategy updateStrategy
|
||||
}
|
||||
|
||||
/** Event marking a state transition in job update lifecycle. */
|
||||
|
|
Loading…
Add table
Add a link
Reference in a new issue